Anthropic CEO Admits Compromising with Authoritarian Regimes to Obtain AI Funding

AIbase基地

Published in AI News · 4 minute read · Jul 23, 2025

In recent internal communications, Dario Amodei, CEO of Anthropic, candidly stated that the company is compromising with some authoritarian regimes to remain competitive in the field of artificial intelligence. This information was initially obtained by WIRED and has sparked widespread attention.

Anthropic, Claude

Amodei said in a Slack message: "It is very difficult for this principle to serve as a guiding principle for business operations: 'No one should benefit from our success.' He felt uneasy about it and further pointed out that Anthropic will seek investments from the United Arab Emirates and Qatar, even if this means benefiting some dictators. He stated that democratic countries need to set strong terms to avoid being overwhelmed by powerful artificial intelligence and to prevent human rights violations by these countries.

When explaining this policy shift, Amodei pointed out that the amount of capital from the Middle East is extremely large, estimated to exceed $100 billion. He said, "If we want to stay at the forefront, accessing this capital would provide us with significant benefits. Without this capital, it would be difficult for us to remain competitive at the forefront."

This statement shows that in the context of rapid development of artificial intelligence technology, obtaining funding is crucial for companies, especially in an increasingly competitive market environment. Although Amodei feels frustrated about such compromises, he seems to realize that it is, in some way, a necessary step for the company's survival and development.

As artificial intelligence technology continues to advance, major companies are seeking investments to accelerate their research and innovation. However, this has also sparked widespread discussions about ethics and social responsibility. Anthropic's decision may have a profound impact on its future development and company image.

Key points:
🌍 Amodei admitted that the company is compromising with authoritarian regimes to obtain financial support.
💰 Anthropic will seek investments from Arab countries, even if this could lead to dictators benefiting.
⚖️ This move has sparked wide discussions about ethics and social responsibility, and the future development is closely watched.

Meta Launches Innovative Model AU-Nets to Revolutionize Text Processing

The Meta research team introduced an innovative architecture called AU-Net, breaking through the limitations of traditional tokenization techniques. The model learns directly from bytes, dynamically combining bytes into hierarchical sequence representations through a self-regressive U-Net structure. Its contracting path compresses bytes into semantic units, while the expanding path restores information and fuses details, with skip connections preserving local features. A multi-stage pooling strategy is used to extract semantics at the word and phrase levels, and multi-linear upsampling optimizes information fusion. The self-regressive generation mechanism ensures text coherence and significantly improves inference efficiency. This architecture

Kimi K2 Technical Report Released! How Does This Model Dominate the Global Open-Source Competition?

The Kimi team released the Kimi K2 technical report. This trillion-parameter large model, with 32 billion activated parameters, won the global open-source model competition, with performance comparable to top closed-source models such as GPT4.5. Its innovations include: using the MuonClip optimizer to process 15.5 trillion token data without loss; developing a multi-domain tool data synthesis pipeline; pioneering the restatement method to improve training efficiency, achieving the effect of ten rounds of original data in one round of rewritten data; optimizing model performance through supervised fine-tuning and reinforcement learning. The model is based on N

Google Releases Stable Version of New Gemini 2.5 Flash-Lite: The Perfect Balance of Speed and Cost

Google released the stable version of the Gemini 2.5 Flash-Lite AI model, focusing on high cost-effectiveness. This model supports a context of up to 1 million tokens, with input/output token pricing at $0.10/$0.40 respectively. The cost for audio input has been reduced by 40% compared to the preview version. Performance exceeds the 2.0 version, featuring functions such as search integration and code execution. Developers are required to migrate to the new version by August 25th, after which the preview version will be discontinued.

OpenAI Wins Trademark Lawsuit to Block Competitor from Using Similar Name

OpenAI won a trademark lawsuit against Open AI, with the court ruling that the latter engaged in misleading practices and ordered it to stop using the similar name and domain open.ai. The ruling stated that Open AI intended to confuse consumers and create a false association with the manufacturer of ChatGPT. This decision not only strengthens OpenAI's market position but also sets an important precedent for trademark protection in the technology industry. OpenAI stated it will continue to protect its brand rights to ensure consumers can clearly distinguish its products. The case highlights the importance of brand protection for businesses.

Breaking Tradition! Latent Labs Launches Web-based AI Model to Democratize Protein Design

Latent Labs launches a revolutionary AI model called LatentX, focusing on innovative protein design. The model, developed under the leadership of Simon Kohl, a former DeepMind core member, can generate new protein structures with atomic-level accuracy and high experimental feasibility. Unlike prediction tools like AlphaFold, LatentX has the ability to creatively design proteins. The company uses a licensing model, offering services to academic institutions and pharmaceutical companies, supported by $50 million in funding. The future plans include launching premium paid features.

21-year-old MIT dropout AI startup team secures $32 million in funding

MIT dropouts' AI compliance firm Delve raised $32M Series A at $300M valuation. Its AI agents automate compliance, aiding rapid client growth from 100 to 500 in 6 months. Tech enables auto-evidence collection, reporting, and tool integration. Plans cybersecurity expansion to maintain lead amid competition. Investors see compliance modernization driving operational upgrades.....